We present a new algorithm for nonlinear semidefinite programming, based on the iterative solution in the primal and dual variables of Karush-KuhnTucker optimality conditions, which generates a feasible decreasing sequence. At each iteration, two linear systems with the same matrix are solved to compute a feasible descent direction and then an inexact line search is performed in order to determ...

Globalized inexact Newton methods are well suited for solving large-scale systems of nonlinear equations. When combined with a Krylov iterative method, an explicit Jacobian is never needed, and the resulting matrix-free Newton–Krylov method greatly simplifies application of the method to complex problems. Despite asymptotically superlinear rates of convergence, the overall efficiency of a Newto...

We present a parallel Newton-Krylov algorithm for solving the three-dimensional Euler equations on multi-block structured meshes. The Euler equations are discretized on each block independently using second-order accurate summation-by-parts operators and scalar numerical dissipation. Boundary conditions are imposed and block interfaces are coupled using simultaneous approximation terms (SATs). ...

Motivated by a problem in large scale climate data analysis, we consider the problem of maximum a posteriori (MAP) inference in graphical models with millions of nodes. While progress has been made in recent years, existing MAP inference algorithms are inherently sequential and hence do not scale well. In this paper, we present a parallel MAP inference algorithm called KL-ADM based on two ideas...

Software metrics are used to measure the quality of a software system. Such metrics indicate the level of desired quality present in a system. However software metrics have traditionally been captured at compile time, rendering useful results, but often times inexact, as the complete source code differs from the executing subset. For this reason, static metrics can fall short of measuring the t...

A network used to detect and recognize several diierent symbols (doors, windows...) in scanned architectural drawings is presented. This network, based on Messmer's network for exact and inexact graph matching, presents a compact representation of all the symbols, which allows a one-pass search. Some modiications to this method for our speciic document analysis problem are outlined: the symbols...

In this paper we propose a parallel implementation of the FSAI preconditioner to accelerate the PCG method in the solution of symmetric positive definite linear systems of very large size. This preconditioner is used as building block for the construction of an indefinite Inexact Constraint Preconditioner (ICP) for saddle point-type linear systems arising from Finite Element (FE) discretization...

Neural networks on chip have found some niche areas of applications, ranging from massive consumer products requiring small costs to real-time systems requiring real time response. Speaking about latter, iterative logarithmic multipliers show a great potential in increasing performance of the hardware neural networks. By relatively reducing the size of the multiplication circuit, the concurrenc...

We present two matrix-free methods for solving exact penalty subproblems on product sets that arise when solving large-scale optimization problems. The first approach is a novel iterative reweighting algorithm (IRWA), which iteratively minimizes quadratic models of relaxed subproblems while automatically updating a relaxation vector. The second approach is based on alternating direction augment...

In this paper we consider the problem of minimizing a smooth function by using the Adaptive Cubic Regularized (ARC) framework. We focus on the computation of the trial step as a suitable approximate minimizer of the cubic model and discuss the use of matrix-free iterative methods. Our approach is alternative to the implementation proposed in the original version of ARC, involving a linear algeb...